Improving trust in science

There’s a letter in the Guardian today calling for study preregistration. This is super exciting - I’ve been wanting this for years.

They’re looking for additional support from past and current journal editors. I am, obviously, not even close to a journal editor (and neither are most of this blog’s readers, I’d wager.) But I’ve passed it along to a bunch of my former professors and PIs. And I’m posting it here. Which I guess is as much as I can do.

If I were to name a holy trinity for improving scientific research, it would be open access, study preregistration, and machine-and-human-readable sharing of the raw data behind experiments.

Open access facilitates review and critique by peers while also increasing the diversity and clarity of the literature itself, by allowing anyone - regardless of affiliation or socioeconomic status - to read, comment upon, translate, and even extend research.

Study preregistration holds researchers accountable for their experimental decisions and sharply decreases degrees of freedom, while also encouraging journals to focus on the quality of the methods rather than the sexiness of the results.

Sharing of raw data allows for more and better meta-analysis, which is an utterly vital part of research, especially in medicine. There are tens of millions of articles in the research literature, yet only a small percentage report enough information to enable meta-analysis. I’ve read studies - recent studies - that report null results with the simple phrase “was not significant”. The loss of this data is a tragedy.

We’re making small steps on each of these points, but we still have a long way to go.